Editor’s note: All due respect to Bob Pease with respect to the title of this blog.
The noise floor is not due to squeaky trusses nor is it about babies crawling around. The noise floor is a term by which signal quality is measured. The noise floor involves a number of factors that are beyond the scope of a blog. However, a simple introduction shows what the noise floor principles are and how technology currently relies heavily on synchronizing to a noise-free clock.
So why is this a factor in relation to society? Well as it turns out, we are a society that lives by clocking. Signals and computers synchronize to a clock. The faster the clock, the more the number of data bits that get transmitted. Although noise was once predominately an analog signal factor, the effect that noise has on clocking is now a driving force in technology. It is here that the noise issue transfers from analog to digital. Noise causes jitter in clocks. Jitter causes clocks to have synchronization issues. Synchronization issues slow down data rates.
Bit Error Rate (BER) is tied to noise. This further slows communication technology. The higher the BER, the more signal is needed to compensate. This includes adding additional check bits or sending signals for comparison all of which chews up bandwidth that could be used for sending data instead of verifying sent data.
The noise floor is similar to the best goal line defense in football which is: don’t let them down there to begin with. In other words, don’t create the noise in the first place and you won’t have to deal with it. In order to sidestep noise, clocking relies on Phase Locked Loop (PLL) technology which is supposed to track the very low noise of crystal generated oscillators, albeit large loop filtering negates that goal. Thus current PLLs are a counterproductive solution, actually introducing much greater loop noise (jitter peaking and jitter accumulation) than that coming from a crystal clock.
PLLs can produce rapid clock signals that are used to synchronize incoming data. Unfortunately, most PLLs are unstable as was shown by Lee in 2003. Couple that with the use of charge pumps in PLLs and you start to see the problem is twofold. The charge pump is creating noise, along with other loop sources of noise (like device noise in VCO), and the PLL is unstable thus resulting in a bandwidth limit. The noise is often filtered further limiting bandwidth. Thus we are clocking to levels that are low due to inadequate technology no matter how fast manufacturers claim their PLLs are.
The various stages of a PLL contribute to the noise floor. Texas Instruments explains this in a subsection titled Noise Sources in PLL in their Technical Brief. These factors include thermal contributions due to excitation of electrons in semiconductors from heating.
A new technology is showing some promise in reducing noise. As shown in the Figures, noise contribution is diminished.
This new technology focuses on instantaneous synchronization with the clock. The clock can be the lowest jitter on a crystal clock or the modulated data on the reference instead of the phase like a PLL uses. The instantaneous lock technique exceeds that of PLLs which have a lower bandwidth and thus cannot perform at the higher clock rates. The results of this technology are quite astounding. Signals exhibit much less noise when compared to a standard PLL. For the first time since introducing the PLL in the 1930’s, synchronization can be on the clock and not the phase.
Furthering this advantage is the reduction in BER which is 70,000 times less. This is not an incremental change. Rather, it’s a disruptive increase in performance as shown by the IW curve in the figure.
The possibilities for the improvements using this technology are far and wide. GPS could be accessed in many more places eliminating dead spots. The same goes for cell coverage. RFID could go from three feet to inventorying an entire row in Walmart. Power reductions would be realized especially in phones that must burn a lot of power to amplify weak GPS signals. Upload speeds would improve by orders of magnitude. Far fewer phone calls would be dropped and Wi-Fi retransmissions would reduce to negligible. The crowded spectrum could be reduced thus avoiding military jam ups. That’s just a part of the benefits. Clocks would go so fast that they couldn’t be intercepted or jammed by adversaries. Military, take note. Quit ignoring the possibilities and start understanding the strategic value.
Of course I’m just an analog engineer making noise about eliminating noise. Whether or not this technology improves is based on money. Once industry realizes that stuffing more software on an already grid locked highway is not the answer, maybe they’ll invest in cars that can go one thousand miles an hour instead of five. I doubt it. All hardware improvements have already been invented and the future is in software. Yeah right, and digital phones will never catch on as per Motorola.
- “Jitter Transfer Characteristics of Delay-Locked Loops—Theories and Design Techniques”
- “Fractional/Integer-N PLL Basics”
- “Thermal Noise Calculator Formulae & Equations,”- formulae & calculations associated with RF or thermal noise and a simple to use voltage and dBm calculator.
- A new technology
- “dBc” Wikipedia explanation of decibels per carrier.